Eur. J. Biochem. 206,697- 704 (1992) 0FEBS 1992

Purification and biochemical characterization of a putative [6Fe-6S] prismane-cluster-containing protein from Desulfovibrio vulgaris (Hildenborough) Antonio J. PIERIK

Ronnie B. G. WOLBERT', Peter H. A. MUTSAERS', Wilfred R. HAGEN' and Cees VEEGER'

Department of Biochemistry, Agricultural University, Wageningen, The Netherlands Cyclotron Laboratory, Eindhoven University of Technology, The Netherlands (Received January 27,1992) - EJB 92 0099

A novel iron-sulfur protein has been isolated from the sulfate-reducing bacterium Desulfovibrio vulgaris (Hildenborough). It is a stable monomeric protein, which has a molecular mass of 52 kDa, as determined by sedimentation-equilibrium centrifugation. Analysis of the metal and acid-labile sulfur content of the protein revealed the presence of 6.3 rfr 0.4 Fe/polypeptide and 6.2 & 0.7 S2-/ polypeptide. Non-iron transition metals, heme, fiavin and selenium were absent. Combining these data with the observation of a very anisotropic S = 1/2 [6Fe-6SI3+prismane-like EPR signal in the dithionite-reduced protein, we believe that we have encountered the first example of a prismanecluster-containing protein. The prismane protein has a slightly acidic amino acid composition and isoelectric point (PI = 4.9). The ultraviolet/visible spectrum is relatively featureless ( E =~ 81 mM-' . cm-l, E~~~ = 25 mM-l . cm-', E ~ = 14~ mM-'~ . cm-I). , The ~ shape ~ of the ~ protein is approximately globular (s20,w= 4.18 S). The N-terminal amino acid sequence is MFSlcFQSic QETAKNTG. Polyclonal antibodies against the protein were raised. Cytoplasmic localization was inferred from subcellular fractionation studies. Cross-reactivity of antibodies against this protein indicated the occurrence of a similar protein in D . vulgaris (Monticello) and Desulfovibrio desulfuricans (ATCC 27774). We have not yet identified a physiological function for the prismane protein despite trials for some relevant enzyme activities.

~

~

Iron-sulfur clusters constitute the redox-active site of a reactions can be accomplished by iron-sulfur clusters [4]. The large number of biological electron-transfer components. The best-characterized example is aconitase, for which an X-ray ubiquity of these versatile redox centers is documented by crystallographical structure is available [51. Unfortunately, the basic set of structures and their respecttheir involvement in vital steps of biochemical pathways: the mitochondria1respiratory chain, photosynthesis, Krebs' cycle, ive spectroscopic properties is not sufficient to explain the methanogenesis, nitrogen fixation, sulfate and nitrate re- EPR characteristics of multielectron redox proteins [6]. duction [l]. The knowledge of iron-sulfur clusters in biological Straightforward elucidation of the structure of the iron-sulfur components in general is based on spectroscopically and crys- sites is hampered by the lack of sufficient resolution of the tallographically scrutinized electron-transfer proteins. Our crystallographical structures, the absence of amino acid sereference collection of X-ray crystallographical structures quence data, and/or the lability of many iron-sulfur centres in is formed by the [2Fe-2S]dinuclear site in ferredoxin, the [4Fe- the aerobic environment [7, 81. As a working hypothesis for 4S] cubane in high-phosphate iron protein, ferredoxin and the description of these structures and their magnetism, we double-cubane ferredoxin, the [3Fe-4S] cluster in ferredoxin have recently proposed the supercluster/superspin concept [9, and [3Fe-4S]/[4Fe-4S]ferredoxin [2]. A number of inorganic lo]. This hypothesis explains the absence of classical ironmodel compounds mimicking these polypeptide-liganded sulfur clusters and the presence of unusual EPR spectra in iron-sulfur cores have been synthesized and characterized (re- multielectron redox proteins with a high Fe/S content by proposing the existence of superclusters of more than four viewed in [3]). Not only electron transport and oxidoreduction, but also magnetically coupled Fe ions, giving rise to unprecedented iron-based substrate coordination and catalysis of non-redox superspin ( S 2 712) paramagnetism. This concept particularly addresses the high Fe/S content of complex enzymes such as Correspondence to W. R. Hagen, Laboratorium voor Biochemie, hydrogenase [l11, nitrogenase component 1 [12], dissimilatory Landbouwuniversiteit, Dreijenlaan 3, NL-6703 HA Wageningen, The sulfite reductase [13] and carbonmonoxide dehydrogenase Netherlands [14]. The existence of larger iron-sulfur clusters, e.g. the [6FeAbbreviation. APS, adenosine 5'-phosphosulfate. 6S] prismane core, is also documented by inorganic model Enzymes. Desulfoviridin or dissimilatory sulfite reductase (EC 1.8.99.1); carbonmonoxide dehydrogenase or acetyl-CoA synthase compounds [15 - 171. In a previous publication, we reported (EC 1.2.99.2); nitrogenase (EC 1.18.6.1); Fe-hydrogenase (EC that an unusual iron-sulfur protein of the sulfate-reducing 1.18.99.1). bacterium Desulfovibrio vulgaris (Hildenborough) exhibited

an EPR signal reminiscent of the [6Fe-6SI3+prismane cluster, while EPR signals characteristic of classical iron-sulfur clusters were absent [18]. Preliminar results on the redox properties and the S = 912 superspin EPR signals of this protein have been communicated [9, 101. We present a biochemical and biophysical characterization of this prismane protein in this and in a subsequent paper [19].

MATERIALS AND METHODS Growth and isolation Desuljovihrio vulgaris (Hildenborough) NCIB 8303 was maintained on Postgate's medium [20]. For localization experiments of Desulfovibrio strains, cells were grown in Saunders' N medium [21] to an absorbance of approximately 0.8 at 660 nm. Large-scale growth of D . vulgaris (Hildenborough) for isolation of the protein was in modified Saunders' medium [22]. A 1YOinoculum was used to start 240-1 batch cultures in a home-built fermentor. Anaerobiosis was obtained by vigorous bubbling with nitrogen gas (99.99%). A 60-h growth period at 35 "C allowed the isolation of 150- 200 g wet cells. Cclls were harvested with a Sharpless continuous-flow centrifuge. The cell paste was suspended in 3 vol. 10 mM Tris/Cl, pH 8.0, in a chilled 1-1Waring blender (10 strokes). Cells were disrupted by three passages through a chilled Manton-Gaulin press (84 MPa). A spatula of DNase and RNase (Sigma) was added. A supernatant containing soluble proteins was separated from a blackish precipitate and dark-red membranes by successive centrifugation steps at 10000 x g (20 min) and 100000 x g (60 min), respectively. The pH of the supernatant was adjusted to 8.0 with 1 M Tris/Cl, pH 9, and diluted with demineralized water to yield a solution with a conductivity of 2mS/cm. The soluble protein fraction was applied onto a DEAE-Sephacel column (5 cm x 20 cm), equilibrated with 10 mM Tris/Cl, pH 8.0. Cytochromes were eluted with starting buffer. After washing with 1 1 10 mM Tris/Cl, pH 8.0, plus 20 mM NaCl, a 20-400 mM NaCl gradient in 10 mM Tris/Cl, pH 8.0, (2 1) was applied to separate the bound proteins into two main fractions: a brownish fraction (60110 mM NaCl) with the prismane protein (see below), (partially inactivated) periplasmic hydrogenase, adenosine-5'phosphosulfate (APS) reductase, assimilatory sulfite reductase and a greenish-brown fraction containing desulfoviridin and ferredoxin. The brownish fraction (typically 400 ml) was concentrated using an Amicon device with YM30 filter. This concentrate was applied in two 10-ml batches onto a 1.5 cm x 100 cm Sephadex G-150 gel-filtration column equilibrated with 25 mM potassium phosphate, pH 7.5. A brown fraction containing the prismane protein eluted at approximately 330 ml and was resolved from a yellow/brown APS reductase fraction. The prismane-protein-containing fractions were passed through a 2 cm x 10 cm Ultragel hydroxyapatite column equilibrated with 25 mM potassium phosphate, pH 7.5, to remove periplasmic hydrogenase and assimilatory sulfite reductase. The eluate was combined with a 50-ml 25 mM potassium phosphate, pH 7.5, wash of the column and was concentrated (Amicon/YM-30). After desalting on Sephadex G-25 (20 mM Tris/CI, pH 8.0) and microfiltration, proteins were applied onto a MonoQ 515 anion-exchange chromatography column attached to a Pharmacia FPLC system. A 60-ml 0 - 1 M NaCl gradient in 20 mM Tris/CI, pH 8.0, resolved the prismane protein eluting at approximately 130 mM NaCl from several other proteins which eluted at higher NaCl concentrations.

Purification, as described, was performed aerobically at 4"C, FPLC at ambient temperature (18 -22°C). Minor precipitates formed during the successive stages of purification were removed by centrifugation at 20000 x g (15 min). During preliminary studies, Western blots (see immunological techniques) were used for the identification of fractions eluting from the first two columns. Later purifications were followed by tracing coeluting enzymes: APS reductase (DEAESephacel) and periplasmic hydrogenase (Sephadex G-150). The final part of the purification was routinely judged by SDS/PAGE with Coomassie-blue staining (see electrophoresis). Molecular mass determination A Pharmacia FPLC system equipped with a Superose 12 (HR 10/30) was used for gel filtration. The column was equilibrated with 50 mM potassium phosphate/l50 mM KCI (PH 7.2). Molecular mass markers were a-chymotrypsinogen (23.7 kDa), ovalbumin (43.5 kDa), D. vulgaris (Hildenborough) periplasmic hydrogenase (57 kDa including Fe and Sz- contents), bovine serum albumin (67 kDa), bovine liver catalase (248 kDa) and horse spleen apoferritin (z467 kDa). For gel filtration under denaturing conditions, 6 M guanidinium hydrochloride/O.1 M Tris/O.5 mM NazEDTA adjusted to pH 8.6 (HCI) was used. The prismane protein and markers (chymotrypsinogen, ovalbumin, bovine serum albumin and a/P subunits of D . vulgaris (Hildenborough) periplasmic hydrogenase) were iodoacetylated prior to gel filtration. Protein solution (2 - 4 mg/ml) in elution buffer was incubated with 2 mM dithiothreitol at 80°C for 5 min. Subsequently, sulfhydryl groups were blocked by treatment with iodoacetic acid (20mM) at room temperature in the dark (60 min). Sedimentation velocity and sedimentation equilibrium measurements were made with an MSE analytical ultracentrifuge with scanning optics. Prismane protein samples were equilibrated with 100 mM potassium phosphate, pH 7.50/ 0.02% NaN3 on a 1 cm x I 0 cm Sephadex G-25 column. Data were analyzed according to Filipiak et al. [23].

Electrophoresis

Tris/glycine based gels (cf. [24]) were cast and run either in a home-built electrophoresis system or with the use of a LKB Midget gel-electrophoresis system. For native gel electrophoresis, SDS was omitted. The composition (mass/ vol.) of the stacking gel was 4% acrylamide and 0.1% bisacrylamide; running gels were 10- 20% acrylamide and 0.4 - 0.07% bisacrylamide. Molecular masses were estimated with the Pharmacia 14.4-94-kDa marker kit. Gel scanning was performed on a LKB Ultroscan XL system equipped with a HeNe laser (633 nm). Flat-bed isoelectric focussing on Serva Precotes (pI5-7) was performed on a LKB Ultrophor electrophoresis unit at 4°C. Markers were hen egg trypsin inhibitor, superoxide dismutase, carbonic anhydrase (both from bovine erythrocytes) and ribonuclease A (bovine pancreas), with isoelectric points at 4°C of 4.6, 5.3, 5.8 and 8.0, respectively [25]. Immunological techniques

The prismane protein (500 pg) was subjected to preparative SDS/PAGE (30 x 14 x 0.15 cm). Protein was detected by precipitation of the SDS by addition of 100 mM KCI [26].

699 After excision and electroelution (ISCO system), 100-pg amounts were mixed with Freund’s complete adjuvant and injected subcutaneously in male New Zealand white rabbits. BalbC mice were injected with the 10-pg amounts. Boosts of antigen in Freund’s incomplete adjuvant were administered three-weekly. Serum obtained after bleeding was used without further purification. For immunoblotting, SDS/PAGE-separated proteins were transferred onto nitrocellulose (Schleicher and Schiill, 0.45 pm) [27]. Goat anti-(rabbit IgG) antibodies or goat anti(mouse IgG) antibodies conjugated to alkaline phosphatase (Bio-Rad, Richmond, USA and Promega Biotec, Madison, USA, respectively) were used as secondary antibodies for immunostaining. Subcellular localization Cells were harvested by centrifugation at 5000 x g (20 min). After suspension in demineralized water (0°C) 1 vol. 100 mM Na2EDTA/100 mM Tris/Cl, pH 9, was added. Incubation at 32°C for 20- 30 rnin and subsequent centrifugation at 5000 x g (5 min) released an orange/red periplasmic fraction (cf. [22]).Spheroplasts were resuspended in 10 mM Tris/C1, pH 8, and disrupted by two passages through a chilled 10-ml French pressure cell. The resulting spheroplast lysate was spun at 5000 x g ( 5 min) to remove cell debris. Finally, centrifugation at 100000 x g (60 min) yielded a greenish cytoplasmic supernatant. The reddish membrane pellet was suspended in 10 mM Tris/Cl, pH 8.0, and spun at 100000 x g (60 min). The pellet was resuspended for further measurements. The following assays were performed to probe the quality of the fractionation : hydrogenase activity was measured by a manometric hydrogen-production assay [22], APS reductase by the AMP/SOj - ferricyanide-reduction assay [28], and desulfoviridin was estimated from the visible spectrum [A630( A + ~~~ ~ ~ / 2 ~) 1 [ 2 9 1 . Ultraviolet/visiblespectroscopy Spectra were recorded on a DW-2000 spectrophotometer interfaced with an IBM computer. Measurements of absorbances at single wavelengths for the determination of absorption coefficients and chemical analysis were made with a Zeiss M4QIIl spectrophotometer with a PI-2 logarithmic converter. Amino acid and N-terminal analysis

Prior to amino acid analysis protein samples were desalted on Sephadex G-25. For the determination of cysteine and methionine, protein lyophilates were treated with performic acid and HBr [30] and evaporated to dryness. Hydrolysis was carried out for 24 h at 110°C (6 M HCI). Minor corrections were made for the degradation of labile amino acids as estimated from time-dependent recoveries of control protein samples. The content of tyrosine and tryptophan was determined by comparison of ultraviolet spectra of trichloroaceticacid-precipitated protein dissolved in 0.10 M NaOH with spectra of tyrosineltryptophan mixtures [31]. Tryptophan was also estimated fluorimetrically [32]. Standard addition of free tryptophan was used to correct for minor quenching. Bovine serum albumin and periplasmic hydrogenase served as appropriate standards, containing two [33] and six [34] tryptophan residues, respectively.

N-terminal analysis of protein samples blotted onto an Immobilon-P (Millipore) support was carried out by gasphase sequencing (Dr Amons, Leiden University, The Netherlands). Chemical analysis Protein was determined with the microBiuret method at 330 nm (351 after trichloroacetic acid/deoxycholate precipitation [36]. Fatty-acid-free bovine serum albumin (Sigma) served as standard using = 6.67 at 279 nm [37]. Careful checks for both standard and sample were made, including centrifugation, omission of precipitation, recording of spectra and blank determinations without copper reagent. Iron was determined colorimetrically after treatment with 1% (mass/vol.) HCl at 80°C for 20 min. The pH of the resulting reaction mixture (500 pl) was adjusted to 5.0 +_ 0.2 with 250 pl 0.4 M NH4C1. Addition of 25 pl 10% (mass/vol.) SDS was followed by reduction of Fe3+ with 50 pl freshly prepared 0.1 M ascorbic acid solution. The mixture was vortexed and the color reaction was started by the addition of 25 pl iron chelator (25 mM). Bathophenanthrolinedisulfonate [38] or ferene [39] was used. Mohr’s salt [(NH4)2Fe(S04)2. 5 H 2 0 ; Merck p.a.1 was used as a standard. Acid-labile sulfur was determined aerobically with the methylene-blue method [40]. Care was taken to prevent loss or oxidation of sulfide. Solutions were pipetted gently and a low surfxe/volume ratio was maintained, except after the FeCl, addition. As protein quenched the formation of methylene blue from iron-sulfur proteins in our experiments, we used standard addition of a freshly prepared anaerobic 1 - 2 mM Na2S solution in 10 mM NaOH. Sodium sulfide crystals (Na2S . 9 H 20, Merck p.a.) were rinsed with demineralized water and thoroughly dried with Whatman paper. Titration of the sulfide standard with iodine/thiosulfate [42]confirmed the titre of the gravimetrically prepared solution. Molybdenum was determined with a microadaptation of the dithiol method [42]. Ammonium heptamolybdate (Merck p. a.), Azotobacter vinelandii nitrogenase component 1 (kindly donated by Dr H. Haaker) and cow milk xanthine oxidase (Boehringer) were standards. Elemental analysis was performed by particle-induced X-ray emission at the Eindhoven University of Technology, The Netherlands. Samples were diluted with ethanol to 20% (by vol.) ethanol, prior to application onto Millipore MF SCWP 8-pm filters (cf. [43]). The filters were left to dry in air. Typical samples contained approximately 0.5 mg protein/ 5 cm’. Standards contained 0.05-5 pg element/5 cm’. The filters, mounted in commercial slide frames, were irradiated with a 3-MeV proton beam from a cyclotron. Energy dispersive X-ray spectra were digitally recorded and analyzed with the software as described in [44]. Materials DEAE-Sephacel and Sephadex G-150 were from Pharmacia, Bio-Gel HTP from Bio-Rad, SDS from BDH Biochemicals and iron chelators from Aldrich. Other chemicals were obtained from Merck. RESULTS AND DISCUSSION Purification While assessing the purity of samples of D. vulgaris (Hildenborough) periplasmic hydrogenase, we noted a con-

700 Table 1. Purification of D. vu@ris (Hildenborougb) prismane protein. The cell-free extract was prepared from 213 g freshly harvested wet cells according to Materials and Methods. Fraction

Volume

Protein

Extract

ml 655

mg 11 500 2130 1970 397

DEAE-Sephacel YM-30 concentrate Sephadex G-350

220 24

Hydroxyapatite FPLC MonoQ

110 1

60

.o

201 5.2

Fig. 1. Homogeneity of D. vulgaris (Hildenborough) prismane protein as revealed by SDS/PACE (A) and IEF (B). Electrophoresis of FPLCpurified prismane protein (lanes 2 in A and B, 1 pg; lane 3 in A, 10 pg) and marker mixtures (other lanes) was carried out according to Materials and Methods.

taminant protein with a molecular mass of 59 kDa, as determined by SDS/PAGE. Subsequent purification of this protein by anion-exchange FPLC of side-fractions from a hydrogenase purification yielded 0.5 mg brownish protein. To obtain a useful tool for characterization and purification of this protein, antibodies were raised against the polypeptide. Pilot purifications of this protein from D . vulgaris cells, monitored by Western blotting with immunodetection, allowed us to purify a few milligrams of the protein. A preliminar EPR characterization [18] pointed towards the presence of a [6Fe6SI3 prismane cluster in the dithionite-reduced protein. This assignment is corroborated by analytical and spectroscopic results presented below and in a subsequent paper [19]. The routine purification starting from about 200 g wet cells provides about 5 mg prismane protein (Table 1). Assuming 50% recovery for the purification, about 0.1 % soluble cellular protein of D. vulgaris (Hildenborough) is prismane protein. This purification yields preparations of 80 -90% purity, as estimated by densitometric scanning of Coomassieblue-stained SDS/PAGE gels. For experiments relying on the homogeneity of the protein the gel-filtration and FPLC anion-exchange steps were repeated at the expense of the yield. The purity of the prismane protein after these additional purification steps was 95 - 99%. Throughout the experiments described in this paper, such preparations were used. +

Homogeneity and molecular mass The prismane protein exhibited a single band on Coomassie-blue-stained SDSjPAGE gels (Fig. 1A). In gels with varying total monomer concentrations of 10-20%,

the apparent molecular mass of the polypeptide was 59.2 5 1.8 kDa. We specifically checked, but found no evidence for, a small subunit, contrary to the cases of D. vulgaris periplasmic hydrogenase [l 11 and dissimilatory sulfite reductase [45]. Native electrophoresis of the prismane protein at pH 9.5 showed a single brown band with a relative mobility of 0.44 with respect to the bromophenol-blue front. Coomassie-blue and iron staining [46] of the native gel demonstrated that the protein comigrated with the iron-sulfur chromophore (not shown). The protein displayed a brown band on flat-bed isoelectric focussing. Close examination of the Coomassie-blue-stained or in situ iron-stained [46] isoelectric-focussing gel revealed that the brown band coincided with two very closely spaced bands (Fig. 1B). The average isoelectric point was 4.9 f 0.1 at 4"C, with a ApI GS 0.03 space between the bands. SDS/PAGE in a second dimension (not shown) proved unequivocally that the polypeptide, iron and brown chromophore were contained in a single entity. Gel filtration of the iodoacetylated denatured prismane protein gives a molecular mass of 62 f 6 kDa. The native molecular mass as estimated by gel filtration (50 mM potassium phosphate/l50 mM KCl, pH 7.2) is 54 f4 kDa. The gelfiltration elution profiles exhibited single symmetrical peaks at 280 nm. To obtain a reliable molecular mass for chemical determinations, in the absence of the complete amino acid sequence, the prismane protein was subjected to short-column Yphantis [47] sedimentation-equilibrium centrifugation. Sedimentation-velocity experiments showed that the protein sedimented as a single boundary, observed by scanning at 280 nm and 400 nm. A szo,w of 4.18 k 0.13 S (n = 4) was calculated from data obtained with solutions of 0.2 - 1 mg prismane protein/ ml100 mM potassium phosphate, pH 7.5, spun at 45000 rpm (20 "C). No significant protein concentration dependence was noted. This value compares with the sedimentation coefficient of the similarly sized D. vulgaris periplasmic hydrogenase, 4.1 S [23]. Sedimentation-equilibrium runs at rotor speeds in the range 11000 - 19000 rpm using identical conditions attained equilibrium after 40 - 80 h. Using a partial specific volume of 0.738 cm3/g, calculated [48] from the amino acid composition (Table 4), a molecular mass of 52.0 f 0.9 kDa (A = 400nm, n = 5 ) and 48.9k 1.5 kDa (A = 280nm, n = 4) was found. SDS/PAGE of samples subjected to a 100-h centrifugation showed no significant breakdown of the prismane protein polypeptide. Background absorbance over the radial axis was negligible at 400 nm. At 280 nm, however, we had to correct for 2 - 10% background absorbance. Also, better fits of ln(A) vs. r2 were obtained with the data taken at 400 nm. We therefore attach greater value to the 400-nm data and use a molecular mass of 52 kDa for quantitative chemical analysis. The sedimentation velocity coefficient and the native gel-filtration behaviour are in agreement with a 52-kDa monomeric structure. Chemical analysis of cofactors Table 2 summarizes the combined results of protein, iron and acid-labile sulfur determinations of several batches of prismane protein. To assess possible interferences of the microBiuret, iron-chelator and methylene-blue colorimetric determinations, we thoroughly checked blanks, calibrations, spectra of color complexes and standard additions. No significant interferences were found for the protein and iron determination. Bathophenanthrolinedisulfonate and ferene chelators

701 Table 2. Iron-sulfur composition and optical absorption coefficients of D. vulgaris (Hildenborough) prismane protein. The protein, iron, and acid-labile sulfur content were determined according to Materials and Methods. Stoichiometries and absorption coefficients were based on determination of the protein concentration with the microBiuret method and calculated using a molecular mass of 52 kDa. n. d., not determined. Preparation 11 was used for ultraviolet/visible spectroscopy in Fig. 2.

Preparation Fe/protein

S2-/protein

E

at

280 nm

1 2 3 4 5 6 7 8 9 10 I1

12 13 Mean t SD (4 a

atoms/molecule

mM-' . cm-'

5.5 6.8 6.4 6.3 6.6 6.5 6.2 n. d. n. d. n.d. 6.9 5.9 n.d. 6.3 0.4 (9)

n. d. n. d. 80.7 : 91.5 80.5 82.5 85.5 84.3 82.9 75.7 82.9 70.0 79.0 81.4 f 5.5 (11)

n.d. n. d. n. d. : 6.7 5.2 6.4 n. d. n.d. 6.4 n.d. n.d. n.d. n.d. 6.2 0.7 (4)

400 nm

n.d. n.d. 26.0 . 27.7 26.8 24.4 21.3 25.5" 25.8 24.6 26.0" 22.7" 23.7" 25.0 f 1.9 (11)

Preparation used for the determination of &,,duced-

Table 3. Physical and chemical properties of D. vulgaris (Hildenborough) prismane protein. _

_

_

_

~

Value

Parameter _

_

_

_

~

Molecular mass S20,W

PI Fe/protein Sz-/protein Mo/protein other transition metals Se/protein eZg0 (isolated) E~~~ (isolated) eaO0 (dithionite reduced)

52.0 f 0.9 kDa 4.18 & 0.13 S 4.9 f0.1 6.3 f 0.4 atoms/molecule 6.2 & 0.7 atoms/molecule < 0.03 atoms/6Fe < 0.05 atoms/6Fe < 0.03 atoms/6Fe 81 f 5 mM-'.cm-' 25.0 & 1.9 mM-' . cm-' 14.1 1.0 mM-' . cm-'

yielded identical iron contents within experimental limits (< 3%). We noted some problems with the methylene-blue method. Recovery of acid-labile sulfur added to protein was 70 -90%. Thus, protein quenched the synthesis of methylene blue from sulfide or, alternatively, methylene blue is adsorbed by the protein [49]. Protein also affected the visible spectrum of the methylene-blue component formed. Methylene blue formed from acid-labile sulfur in the absence of protein exhibited a spectrum identical with authentic methylene blue. However the A670/A750of the product formed in the presence of protein was lower (1.5 instead of 2.0). Beinert [50] briefly mentioned that this ratio is not entirely constant. A tentative explanation is that binding of methylene blue to the denatured polypeptide occurs with a concomitant spectral change of the methylene blue. For quantitative determinations, standard additions of sulfide and the 670-nm absorbance were used.

0.4 1

I

1

I

I

I

I

0.3 Y

Y $ 0.2 r 0 n m a

400

500 600 700 WAVELENGTH (nm)

O 300

400

500 600 WAVELENGTH lnml

700

Fig. 2. Ultraviolet/visible spectroscopy of D. vulgaris (Hildenborough) isolated prismane protein (0.21 mg/ml, 50 mM Hepes, pH 7.5) (A), dithionitereduced (B) and isolated minus dithionite-reduced(inset, A B). Asterisks denote spectral contributions from excess dithionite and its decomposition products.

-

Elemental analysis with proton-induced X-ray emission accurately reproduced the Fe content as determined by the colorimetric technique (three batches). However, the standard deviation for duplicate determinations of the same sample was larger. This can be explained by a minor heterogeneity of the spreading of the samples over the filter (cf. [43]). We therefore used the iron present in the prismane protein samples as an internal standard for the determination of other elements, and relied on the colorimetric method for the iron content. Protoninduced X-ray emission revealed that the prismane protein as isolated (three batches) contained less than 0.05 atoms V, Cr, Mn, Co, Ni, Cu, Se, Mo or W/6 Fe atoms. The relatively high and variable Zn content of the filters (80- 120 ng/cm2) complicated an accurate analysis of the Zn content of the prismane protein. Two preparations contained less than 0.05 atoms Zn/6 Fe atoms and one preparation about 0.1 atom Zn/6 Fe atoms. Since the prismane protein isolated exhibited an EPR signal reminiscent of a Mo(V) center [18], the molybdenum content was cross-checked by a colorimetric procedure. Molybdenum released from xanthine oxidase and nitrogenase component 1 was readily detected, whereas less than 0.03 mol Mo/mol protein was found in the prismane protein. The heme content of the prismane protein was less than 0.005 mol/mol, as estimated from the absence of potential a-band absorbances in the 500 -600-nm region of the (difference) visible absorbance spectrum (Fig. 2). Non-covalently bound FMN and FAD were not detected by fluorescence spectroscopy of acid extracts of the protein. The presence of protein-bound flavin was excluded by ultraviolet/visible spectroscopy of alkaline solutions of trichloroacetic-acid-precipitated protein. Ultraviolet/visiblespectroscopy The absorbance spectrum of the prismane-containingprotein exhibits a broad 400-nm band and a pronounced aromatic peak centered at 280 nm (Fig. 2). The A400/A280ratio of the isolated protein is 0.307 f 0.031 (n = 11). Absorption coefficients at 280 nm and 400 nm are 81.4 +_ 5.5 mM-l . cm-'

702

Fig. 3. Cellular localization of D. vulgaris (Hildenborough) prismane protein by imrnunostaining. Western blots were immunostained after treatment with 500-fold-diluted rabbit antiserum against D. vulgaris (Hildenborough) prismane protein. Lane 1, lysed cells; lane 2, periplasma; lane 3, cytoplasma; lane 4, membranes (each lane equivalent to 80 pg cells);lane 5, purified periplasmic Fe-hydrogenase (200 ng); lane 6, purified prismane protein (10 ng). Arrows indicate the electrophoretic mobility of the prismane protein, as revealed by Coomassie-blue staining after SDS/PAGE. and 25.0 k 1.9 mM-' . cm-', respectively (Table 2). Titration with substoichiometric amounts of potassium ferricyanide resulted in an increase in absorption coefficient at 400 nm of 1.0 mM-' . cm-' in addition to the contribution by potassium ferricyanide. The 400-nm absorption coefficient of the dithionite-reduced protein is 14.1 1.0 mM-' . cm-' (n = 4). Similar results were obtained with hydrogen reduction in the presence of 5 nM catalytic traces of methylviologen/hydrogenase (pH 8.0). The featureless shape of the visible spectrum, both of the isolated protein and the dithionite-reduced protein, compares to that of bacterial ferredoxins containing [4Fe-4S] or 2 x [4Fe-4S] clusters. It clearly has no resemblance to the structured visible spectra of rubredoxin-like and [2Fe-2S] containing proteins.

Fig. 4. Screening for a prismane protein in Desurfovibrio strains by immunostaining. Western blots were immunostained after treatment with 1000-fold-diluted mouse antiserum against D. vulgaris (Hildenborough)prismane protein. Lane 1, purified prismane protein (20 ng); lane 2, cells from D. vulgaris (Hildenborough; NCIB 8303); lane 3, D. gigas (NCIB 9332); lane 4, D . dcsulfuricans (ATCC 27774); lane 5, D. desulfuricans Norway 4 (NCIB 8310); lane 6, D. vulgaris (Monticello) (NCIB 9442); (lanes 2-6, extracts of 500 pg cells/lane). Arrows indicate the electrophoretic mobility of the prismane protein, as revealed by Coomassie-blue staining after SDS/PAGE.

the protein was released from membranes during the process of cell disruption [52], the absence of a relevant immunoblot response of the membrane fraction strongly argues against this possibility. A fully soluble nature of the prismane protein is also indicated by both amino acid composition (Table 4) and aqueous solubility ( > 40 mg/ml). No significant crossreactivity of the antiserum was noted with D . vulgaris (Hildenborough) periplasmic hydrogenase (Fig. 3, lane 5) or the lacZ-hydC fusion product [53](not shown), nor did antiserum against these proteins react with the prismane protein on Western blots (not shown). Mouse sera were also used for immunoblot experiments. No differences were seen.

Cellular localization

Occurrence in other Desulfovibrio strains

Freshly grown D . vulgaris (Hildenborough) cells were fractionated by EDTA extraction, disruption and centrifugation, as described in Materials and Methods. From the EDTA extractability [22] and membrane-translocation mechanism [Sl] it is known that in D . vulgaris (Hildenborough) a highly active Fe-hydrogenase resides in the periplasm, whereas APS reductase and dissimilatory sulfite reductase (desulfoviridin) are present in the cytoplasm [52]. These enzymes were used as convenient markers in our fractionation procedure. In typical localization experiments, about 90% of the respective enzymes in the correct fractions were recovered. The hydrogenase activity of the cytoplasmic fraction ( 5 - 10%) was due to incomplete extraction as was indicated by Western blotting with immunodetection using polyclonal antibodies against the cr-subunit of the periplasmic hydrogenase. A minor contamination with the soluble marker enzymes was found in the membrane fraction. The localization of the prismane protein was studied by immunostaining of Western blots containing fractions equivalent to 80 pg cells (Fig. 3). It is clear that the prismane protein is cytoplasmic. Although it is possible that

lmmunoblots of cell-free extracts of Desulfovibrio strains treated with mouse antiserum exhibited a clear cross-reactive band with a similar mobility [ D . vulgaris (Monticello)], a faint band with similar mobility (Desulfovibrio desulfuricans ATCC 27774) or faint bands with lower mobilities ( D . desuljiuricans Norway 4 and Desulfovibrio gigas; Fig. 4). If we assume the subunit structure and molecular mass of the prismane protein in Desulfovibrio strains to be similar, the presence of a prismane protein in D . vulgaris (Monticello) and D . desuljiuricans (ATCC 27774) is indicated by our data. The presence in the latter species is corroborated by a recent report confirming prismane-protein EPR signals [9, 10, 18, 191 in a protein isolated from D . desulfuricans (ATCC 27774) [54]. Amino acid composition and N-terminal sequence The results of the amino acid analysis of the prismane protein are shown in Table 4. Performic acid oxidation only allowed an rough estimation of cysteine residues. Substances

703 Table 4. Amino acid composition of D. vulgaris (Hildenborough)

prismane protein. Addition of the atomic masses of 6Fe and 6s to the total molecular mass of the consituent amino acids (51.5 kDa) yields the observed native molecular mass (52.0 kDa).

Amino acid

Asx Glx Ser Thr Pro His LYS Arg GlY Ala Val Ile Leu TYr Phe Trp Met CYS Total a

Amino acid composition relative

calculated

mol/mol leucine

mol/mol protein

1.061 0.952 0.487 0.585 0.690 0.190 0.713 0.296 0.995 1.111 0.760 0.531 1 0.290 0.385 0.092 0.106 0.377

48 43 22 27 31 9 32 14 45 50 34 24 45 13 18 4 5 17” 48 1

-

This value probably is an overestimation (see text).

eluting near the cysteic acid peak considerably complicated quantitation and might have caused an overestimation of the number of cysteine residues. The preponderance of acidic over basic amino acid residues is in agreement with the slightly acidic isoelectric point. The N-terminal sequence of the prismane protein was determined with gas-phase sequencing: MFSlcFQS/ ,QETAKNTG. Presumably the N-terminal methionine represents the deformylated initiator methionine of the prismane protein. Comparison of the N-terminal sequence with several protein sequence data banks did not result in significant matches. Physiological function A number of relevant activity measurements [28] were carried out in order to investigate the physiological function of the prismane protein. The hydrogen-producing hydrogenase activity was 0.7 U/g. Fumarate, sulfite, nitrite, thiosulfate and APS reductase activity were less than 5 U/g. Lactate and formate dehydrogenase activity was less than 0.1 U/mg. The reactivity with NADH or NADPH was low; in diaphorase assays with 2,6-dichloroindophenol or horse heart cytochrome c as acceptor the activity was less than 1 U/g. Inactivation as an explanation for the absence of enzyme activity is unlikely. The stability of the protein (as judged by e.g. solubility, intactness of the polypeptide and visible chromophore) on exposure to trypsin or Staphylococcus aweus V8 protease, incubation at room temperature for more than 100 h and repeated dilution/concentration and freezing/ thawing cycles is remarkable. Conclusions The results presented here delineate the biochemical properties of a new type of iron-sulfur protein and provide a

basic set of data necessary for a detailed biophysical study [19]. By an extensive evaluation of the homogeneity and by chemical analysis of protein preparations, the iron and acidlabile sulfur content was pinpointed at about 6.3 atoms/molecule (Tables 2 and 3). If we assume a reasonable relative inaccuracy of 10 - 20% for the combined results of ironlacidlabile-sulfur, protein determination and molecular mass, the Fe/S content is restricted to 5, 6 or 7 atoms/molecule. As above, EPR and Mossbauer spectroscopy [18,19] prove that rubredoxin-like, [2Fe-2S], [3Fe-4S] and [4Fe-4S] centers are absent in the protein, the presence of a single, larger Fe/S (super)cluster is thus indicated. Furthermore, the striking similarity between the EPR g tensors of the [6Fe-6SI3+ model compound and the protein [18] lends additional support to an assignment as prismane protein. However, one should recall that (bi)capped prismane structures exist and might share spectroscopic properties with their prismane parent [16,17]. We anticipate that the 52-kDa prismane protein is not an electron carrier but rather an iron-sulfur redox enzyme. A physiological function (i. e. enzyme activity), however, has not been detected yet. The presence of a [6Fe-6S] supercluster with superspin paramagnetism (see also [19]) points towards a multielectron redox enzyme (EC class 1). We will continue to put efforts in the elucidation of the physiological function of the prismane protein. Determination of the DNA sequence of the prismane-protein gene (presently in progress), sequence comparison and identification of potential flanking genes will stimulate future work. Knowledge of the primary structure would improve the reliability of the Fe/S stoichiometry by substitution of the sedimentation equilibrium molecular mass. We expect that the purity, remarkable stability and homogeneity with respect to charge allows crystallization of the prismane protein. We wish to thank Dr W. M. A. M. van Dongen and Mrs A. Kaan for their kind help with immunological techniques and N-terminal sequencing. We are indebted to Mr L. C. de Folter (Eindhoven University of Technology, The Netherlands) for help with the protoninduced X-ray emission measurements. Mr L. J. G. M. Bongers (Dept of Human and Animal Physiology, Agricultural University, Wagcningen) kindly carried out the amino acid analysis. Mr J. Haas took care of the handling of animals. Mr A. H. Westphal skillfully participated in the ultracentrifugation data analysis. This investigation was supported by the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Organization for Scientific Research (NWO).

REFERENCES 1. Thomson, A. J. (1985) in Metalloproteins (Harrison, P. M., ed.) vol. 1, pp. 79- 120, Verlag Chemie, Weinheim. 2. Fujii, T., Moriyama, H., Takanaka, N., Wakagi, T. & Oshima, T. (1991) J . Biochem. (Tokyo) 110,472-473. 3. You, J.-F., Snyder, B. S., Papaefthymiou, G. C. & Holm, R. H. (1990) J . Am. Chem. SOC.112,1067-1076. 4. Switzer, R. L. (1989) Biofactors 2, 77-86. 5. Beinert, H. & Kennedy, M. C. (1989) Eur. J . Biochem. 186, 515. 6. Hagen, W. R. (1987) in Cytochrome systems (Papa, S . , Chance, B. & Ernster, L., eds) pp. 459-466, Plenum Press, New York. 7. Bolin, J. T., Ronco, A. E., Mortenson, L. E., Morgan, T. V., Williamson, M. & Xuong, N . H. (1990) in Nitrogenfixation: achievements and objectives (Gresshoff, P. M., Roth, L. E., Stacey, G. & Newton, W. E., eds), pp. 117-122, Chapman

and Hall, New York. 8. McRee, D. E., Richardson, D. C., Richardson, J. S. & Siegel, L. M . (1986) J . Biol. Chew. 261, 10277-10281.

704 9. Hagen, W. R., Pierik, A. J. & Veeger, C. (1990) Ital. Biochem. SOC.Trans. 1, 85. 10. Hagen, W. R., Pierik, A. J., Wolbert, R. B. G., Wassink, H., Haaker, H., Veeger, C., Jetten, M. K., Stams, A. J. M. & Zehnder, A. J. B. (1991) Biofactors 3, 144. 11. Hagen, W. R., Van Berkel-Arts, A., Kriise-Wolters, K. M., Dunham, W. R. & Veeger, C. (1986) FEBS Lett. 201, 158162. 12. Hagen, W. R., Wassink, H., Eady, R. R., Smith, B. E. & Haaker, H. (1987) Eur. J. Biochem. 169,457-465. 13. Pierik, A. J. & Hagen, W. R. (1991) Eur. J. Biochem. 195, 505516. 14. Jetten, M. S. M., Pierik, A. J. & Hagen, W. R. (1991) Eur. J. Biochem. 202, 1291- 1297. 15. Kanatzidis, M. G., Hagen, W. R., Dunham, W. R., Lester, R. K. & Coucouvanis, D. (1985) J. Am. Chem. SOC.107, 953961. 16. Coucouvanis, D. (1991) Acc. Chem. Res. 24, 1-8. 17. Noda, I., Snyder, B. S. & Holm, R. H. (1986) Znorg. Chem. 25, 3851 -3853. 18. Hagen, W. R., Pierik, A. J. & Veeger, C. (1989) J. Chem. SOC. Faraday Trans. 185,4083-4090. 19. Pierik, A. J., Hagen, W. R., Dunham, W. R. & Sands, R. H. (1992) Eur. J. Biochem. 206, 705-719. 20. Postgate, J. R. (1951) J. Gen. Microbiol. 5 , 714-724. 21. Saunders, G. F., Campbell, L. L. & Postgate, J. R. (1964) J. Bacteriol. 87, 1073- 1078. 22. Van der Westen, H. M., Mayhew, S. G. & Veeger, C. (1978) FEBS Lett. 86, 122-126, 23. Filipiak, M., Hagen, W. R. & Veeger, C. (1989) Eur. J. Biochem. 185,547-553. 24. Laemmli, U. K. (1970) Nature 227,680-685. 25. Wolbert, R. B. G., Hilhorst, R., Voskuilen, G., Nachtegaal, H., Dekker, M., Van’t Riet, K. & Bijsterbosch, B. H. (1989) Eur. J . Biochem. 184,627-633. 26. Prussak, C. E., Almazan, M. T. & Tseng, B. Y. (1989) Anal. Biochem. 178,233-238. 27. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl Acnd. Sci. USA 76,4350-4354. 28. Odom, J. M. & Peck, H. D. (1981) J. Bacteriol. 147,161 - 169. 29. Aketagawa, J., Kojo, K. & Ishimoto, M. (1985)Agric. Biol. Chem. 49,2359 -2365. 30. Hirs, C. H. W. (1967) Meihodr Enzymol. 11, 59 - 62. 31. Bencze, W. L. & Schmid, K. (1957) Anal. Chem. 29,1193-1196.

32. Pajot, P. (1976) Eur. J. Biochem. 63,263-269. 33. King, T. P. & Spencer, M. (1970) J. Biol. Chem. 245,6134-6148. 34. Voordouw, G. & Brenner, S. (1985) Eur. J. Biochem. 148, 515520. 35. Goa, J. (1953) Scand. J. Clin. Lab. Invest. 5, 218-222. 36. Bensadoun, A. & Weinstein, D. (1976) Anal. Biochem. 70, 241 250. 37. Foster, J. F. & Sterman, M. D. (1956) J. Am. Chem. Sor. 78, 3656 - 3660. 38. Blair, D. & Diehl, H. (1961) Talanta 7, 163- 174. 39. Hennessy, D. J., Reid, G. R., Smith, F. E. & Thompson, S. L. (1984) Can. J . Chem. 62,721 -724. 40. Lovenberg, W., Buchanan, B. B. & Rabinowitz, J. C. (1963) J. Biol. Chem. 238,3899-3913. 41. Vogel, A. I. (1961) A textbook of quantitative inorganic analysis, 3rd edn, pp. 370-371, Longmans, London. 42. Hart, L. I., McGartoll, M.A., Chapman, H. R. & Bray, R. C. (1970) Biochem. J. 116,851-864. 43. Kivits, H. P. M., De Rooij, F. A. J. & Wijnhoven, G. P. J. (1979) Nuclear Instrum. Methods 164, 225 - 229. 44. Johansson, G. I. (1982) X-ray spectrom. 11, 194-200. 45. Pierik, A. J., Duyvis, M. G., Van Helvoort, J. M. L. M., Wolbert, R. B. G. & Hagen, W. R. (1992) Eur. J. Biochem. 205, 111 115. 46. Ohmori, D., Tanaka, Y., Yamakura, F. & Suzuki, K. (1985) Electrophoresis 6, 351 - 352. 47. Yphantis, D. A. (1964) Biochemistry 3, 297-317. 48. Cohn, E. J. & Edsall, J. T. (1943) Proteins, amino acid and peptides, p. 372, Van Nostrand-Reinhold, Princeton, New York. 49. King, T. E. & Moms, R. 0. (1967) Methodr Enzymol. 10,634637. 50. Beinert, H. (1978) Anal. Biochem. 131, 373-378. 51. Van Dongen, W., Hagen, W. R., Van den Berg, W. & Veeger, C. (1988) FEMS Microbiol. Lett. 50, 5-9. 52. Kremer, D. R., Veenhuis, M., Fauque, G., Peck, H. D., LeGall, J., Lampreia, J., Moura, J. J. G. & Hansen, T. A. (1988) Arch. Microbiol. 150,296 - 301. 53. Stokkermans, J., Van Dongen, W., Kaan, A,, Van den Berg, W. & Veeger, C. (1989) FEMS Microbiol. Lutt. 58,217-222. 54. Ravi, N., Moura, I., Tavares, P., LeGall, J., Huynh, B. H. & Moura, J. J. G. (1991) J . Znorg. Biochem. 43,252.